約一百年前，美國到達“馬匹頂峰”。1920年，有近2500萬匹馬穿行于全國的平原、大道、胡同、競技場、飼養場、港口、農場和臟暗的小巷，負責馱運貨物、耕地、打仗，以及拖著馬車搭載乘客，臟亂不堪?！睹绹鴤鹘y》（American Heritage）雜志稱，1900年紐約州羅切斯特市（Rochester, N.Y.）衛生部門的官員估計，全市15000匹馬每年產生的糞便足以堆起方圓一英畝、高175英尺的大糞堆。到1930年，美國的馬匹數減至1900萬。再到1960年，全國只剩下300萬匹馬。馬兒作為昔日主要的交通工具，終于被一項更強勁、更無糞便之憂的新技術徹底取代——“無馬之車”，也就是汽車。

一個世紀后，像通用汽車（General Motors，本年度《財富》美國500強排名第10）和福特（Ford，排名第11）一類的汽車制造商面臨的難題則是，各自的“無馬之車”是否即將赴馬兒的后塵。我們或已到達甚至越過了“汽車頂峰”。美國乘用車成交量于2016年創歷史新高，為1750萬輛，但于2017年跌至1720萬輛。十幾或二十多歲的人群中，考駕照的人少了：1983年，20至24歲人群中注冊比例高達92%，2014年只有77%。

使用汽車不再意味著需要擁有汽車。2017年，Uber全球載客40億乘次。新一代乘客熱衷于拼車——在Lyft提供拼車（Line）功能的城市，40%的乘次由兩名以上乘客共享。而且，從五花八門的發布和鋪天蓋地的報道來看，汽車共享、電動并自動駕駛的科技具有不遜于當年第一批“無馬之車”的前景，其發展方興未艾。

About a hundred years ago, the United States reached “peak horse.” In 1920 some 25 million horses roamed the plains, boulevards, cul-de-sacs, rodeos, stockyards, ports, farms, and dingy alleys of America—toting freight, plowing fields, fighting wars, carrying passengers on buggy rides, and making a complete mess. According to American Heritage magazine, in 1900 health officials in Rochester, N.Y., estimated that the city’s 15,000 horses produced enough manure to make a 175-foot-high, one-acre-round pile every year. By 1930 the American horse population had dropped to 19 million. And by 1960 the country had just 3 million horses. The horse had been fully displaced as the dominant mode of transportation by a new technology that was both more powerful and less prone to produce manure—the horseless carriage, a.k.a., the car.

A century later, the question facing automakers like General Motors (No. 10 on this year’s Fortune 500) and Ford (No. 11) is whether their horseless carriages are about to go the way of the horse. We may well have reached, or even passed, “peak car.” A record 17.5 million passenger vehicles were sold in the U.S. in 2016, but that number dipped to 17.2 million in 2017, and it could fall under 17 million this year. Fewer teens and twenty?somethings get their driver’s licenses: While 92% of 20-to-24-year-olds were registered in 1983, just 77% were in 2014.

Alternatives to ownership are taking off. In 2017, Uber provided 4 billion rides worldwide. The new generation of passengers has embraced ride-sharing—in cities where Lyft has launched its Line service, 40% of its rides are shared by two or more passengers. And gauged by myriad announcements and breathless media coverage, a world of shared electric, autonomous cars—a technology with all the promise of those first “horseless vehicles”—is just around the corner.

毫不夸張地說，對于通用汽車這樣的傳統汽車制造商，汽車銷量的減少似乎帶來了挑戰。近一個世紀以來，通用汽車所為無非造車，然后再把車賣給客戶。未來短期內，公司仍將如此，但長期局面遠不夠明朗。汽車生產商、芯片制造商、拼車網絡運營商和自動駕駛軟件提供方都在不斷變化，交織在一起，使得最終形勢的預測極為困難。高德納（Gartner）調研公司的分析人員邁克·拉姆齊長期關注這一領域，他告訴我：“未知因素太多了?！蓖ㄓ闷嚬緝炔亢屯馊私灾氖牵哼@家110年老牌汽車企業的安穩日子早已不再，未來誰都能坐莊。

再來說說瑪麗·巴拉。她于2014年1月擔任通用汽車公司CEO，成為美國汽車制造界第一位女掌門人。不同于百視達（Blockbuster）和柯達（Kodak）這類此前受災企業的領導人，巴拉不否認當前面臨的劇變?！拔蚁嘈鸥淖??！彼?月上旬一次采訪中對我說。

我們在帕洛阿爾托市（Palo Alto）四季酒店（Four Seasons hotel）一間名為“協同”（Synergy）的樸素的會議室內面對面坐著。酒店矗立于101國道旁，看似一臺巨大的玻璃結構蘋果HomePod智能音箱。而巴拉身披一件黑色皮夾克，笑容或有些許戒備，但也很熱情。她既顯得平易近人，又表現出對通用汽車公司及其員工相當的偏護。

她剛從底特律飛來，要前往曾授予她MBA學位的斯坦福商學院，為勞拉·阿里利亞加－安德森的學生講一堂課。“自動駕駛較之人為駕駛更安全，該技術正在來臨，并且，再加上人工智能和機器學習，前景會越來越好?！彼f道。

巴拉與通用汽車公司有很深的淵源。她就出生在底特律附近，畢業于當時的通用汽車學院（General Motors Institute）——一個為公司提供畢業生人才的校企合作項目，并獲得電機工程學位。她的父親雷·馬戈拉曾在旁蒂克（Pontiac）一間工廠擔任39年的制模工。

A world that buys fewer cars seems to pose a challenge, to say the least, to a traditional auto manufacturer like General Motors. For almost a century, GM has engaged in a single process: building cars and selling them to individuals. It will continue to do that in the short term, but the long term is much less clear. Predicting the ultimate shape of this evolving mashup of auto manufacturers, chipmakers, ride-share network operators, and autonomous software providers is immensely difficult. As Mike Ramsey, an analyst at research firm Gartner who follows this space closely, tells me: “There are a lot of unknowns.” One thing that everyone acknowledges, inside and outside of GM: The 110-year-old automaker’s halcyon days are long gone, and the future is up for grabs.

Enter Mary Barra, who took over as CEO of GM in January 2014, becoming the first woman to run a U.S. automaker. Unlike the leaders of past victims of disruption like Blockbuster and Kodak, Barra isn’t in denial about the radical change that’s afoot. “I believe in it,” she tells me in an interview in early May.

We are seated across from each other in the anodyne “Synergy” conference room of Palo Alto’s Four Seasons hotel, which rises above Route 101 like a great glass Apple HomePod. Barra, sporting a black leather jacket, has a warm if guarded smile. She comes across as both approachable and extremely protective of GM and its people.

She has just flown in from Detroit to speak to a class taught by Laura Arrillaga-Andreessen at Stanford’s B-school, where Barra got her MBA. “Autonomous technology that’s safer than a car with a human driver is coming,” she explains, “and it’s going to get better and better and better with technologies like artificial intelligence and machine learning.”

Barra has GM in her bones. She was born just outside Detroit and graduated with a degree in electrical engineering from what was then called the General Motors Institute, a co-op university program that fed graduates into the company. Her father, Ray M?kel?, was a diemaker who worked for 39 years in a Pontiac factory.

現在她必須讓這家汽車巨頭急劇轉變發展軌跡。她的挑戰是要在維持乃至優化舊模式從而創收的同時，再造通用汽車公司，使之領跑當下疾速演變的出行交通產業。借用該公司高層和管理專家鐘愛的一個詞——通用汽車必須做到“膽大心細”（ambidextrous）。一方面，巴拉必須協助核心業務取得卓越并不斷提升；100%的營業額和利潤都來源于此。但對于邊緣項目，她也必須在通用汽車公司自動駕駛的電動汽車的研發上予以推動并投入資金。此項工作由Cruise Automation的凱爾·沃格特及其500人團隊主導。通用汽車公司于2016年初以10億美元收購了這家舊金山初創企業。

“膽大心細”管理法由斯坦福商學院的查爾斯·奧萊理在《領導與干擾》（Lead and Disrupt）一書中提出，他與哈佛大學的邁克爾·塔什曼合著此書?！拔疫€沒有見過哪個公司因為有了技術而倒閉的。公司要么自己有技術，要么沒技術就去買技術。這是決策的問題。”奧萊理就著早餐說道。奧萊理任教于斯坦福為通用汽車公司培養主管的一個項目，他認為巴拉已成功開啟了通用汽車公司在未來參與競爭所必需的轉型?！霸谖铱磥恚ㄓ闷嚹壳暗淖龇ㄏ噍^于其眾對手更可能取得成功，盡管沒人能夠保證?！彼f。

Now she must prepare the auto giant for a radical lane change. Barra’s challenge is to reinvent General Motors as a leader in the rapidly evolving transportation world, while simultaneously delivering great profits by doing what GM has always done—only better than it ever has. To use a term that is much favored by the company’s leaders and by management gurus, GM must be ambidextrous. On the one hand, Barra must help the core business to excel and continually improve; that’s where 100% of the company’s revenues and profits come from. On the edge, however, she must accelerate and invest in GM’s effort to develop autonomous electric cars, an effort led by Kyle Vogt and the 500 employees of Cruise Automation, the San Francisco startup that GM purchased in early 2016 for around $1 billion.

Charles O’Reilly is a Stanford B-school professor who laid out the concepts of ambidextrous management in Lead and Disrupt, a book he wrote with Harvard’s Michael Tushman. “I’ve yet to find a company that failed because of technology,” O’Reilly explains over breakfast. “They either had it or they could buy it. It’s a leadership issue.” O’Reilly, who teaches in a program Stanford runs for GM execs, believes that Barra has successfully driven changes that GM must make to compete in the future. “I think what GM is doing is more likely than their competitors to be successful,” says O’Reilly. “Although there’s no guarantee.”

兩年前，我第一次采訪瑪麗·巴拉。她對我說：“要改變企業文化，首先必須改變人們的行為。關鍵不在于你說什么，而在于你做什么?！?

當時的通用汽車經歷了一次點火開關工程紕漏事件，涉及一百多人遇難。還在事后的余波中掙扎。巴拉這話在公司內部引起相當共鳴。此前幾十年來，公司因踢皮球的官僚作風而受到詬病。后來調查上述災難事件的起因，就直接聲討了這“與我何關”的作風。巴拉以這次調查為契機，加緊從公司內部發動轉型。而到2016年，效果開始顯現。幾百項內部改革的著力點——統稱為“通用汽車2020”（GM 2020）——開始收獲成效。完成了第一次為期兩年的“轉型領導人”（Transformational Leaders）計劃后，70位主管開發出了一系列新項目，例如汽車共享服務Maven。在密歇根州沃倫市的通用汽車技術中心，一整層樓為Maven騰出，有十多個小隔間，幾間帶門的辦公室，一片鋪著地毯的宣傳和營銷用地，以及許多空余場所。

今年春季，我回沃倫參觀通用汽車技術中心的大幅整修。從底特律市區的通用汽車總部往北半小時車程，就是技術中心，于20世紀50年代由埃羅·沙里寧設計，是通用汽車公司真正的大心臟。約2000名工程師、設計師和IT專家在此處38幢樓里工作。

邁克爾·阿里納是一名工業工程師，曾在麻省理工學院（MIT）學習人際網絡和創新，現在是通用汽車的人才主任。“兩年前你所看到的只是開始，”他說，“現在，這一切已經在公司的血脈里了?！?

阿里納所指的是文化變革，但用來形容我們目前所在的汽車工程中心大樓也很貼切。大堂寬敞，有兩層樓高，通風良好，四周墻面是玻璃。大堂內有一家星巴克，各種舒適的沙發長椅，和電影《警察與卡車強盜》（Smokey and the Bandit）中的一輛特蘭斯艾姆 （Trans Am）。我們在一張極其低調的前臺辦公桌前登記，上了二樓。“都是為了大家的工作，”阿里納說。“為了創建適合大家不同工作方式的空間?！?

針對性格內向的員工和喜好安靜的員工，辦公室里配備了一些 Steelcase公司生產的便攜式Brody辦公隔間。幾排白色桌子用臨時墻隔離出來，墻上有冷灰色和明亮的三原色嵌板，這是團隊任務的場所。還有一處配有一臺大型平板電視的休息區，以及用來鼓勵隨意交流的寬敞開放空間。餐區的桌面由老舊別克和雪佛蘭的引擎蓋鑿成。這一切都自大樓的玻璃外墻往回收，從而留出一道622英尺的長廊，向外就能眺望極具年代感的園區，而樓層的地面則鋪滿了自然光。我一時想起蘋果在庫比蒂諾（Cupertino）新建的總部大樓沿弧形玻璃外墻足足一英里的長廊。

確實，無論在谷歌（Google）、Facebook、赫曼米勒（Herman Miller）、或是其他任何一家我們所熟知的所謂創新企業，沃倫新園區的任何元素都不會顯得格格不入。通用汽車高管走訪所有上述企業的園區，捕捉靈感，以實現與公司全新工作流程相匹配的全新設計。“重點不在于裝修，”巴拉說道，“而在于創造協作環境，并為人們高效工作提供所需資源。我們怎樣才能確保員工所處的工作環境起到賦能和賦權的作用，而不是壓抑和限制才能？”

Two years ago, when I first interviewed Mary Barra, she told me, “To change a culture, you have to change behaviors. It’s not what you say, it’s what you do.”

Back then, those words had a particular resonance at GM, which was still wrestling with the aftermath of an ignition switch engineering flaw that was linked to the deaths of over 100 people. For decades, the company had been known for its pass-the-buck bureaucracy. That “not my problem” behavior was directly implicated in the study GM commissioned to examine the causes of the debacle. Barra used the study as a cudgel to accelerate her efforts to change GM from within, and in 2016 she was just starting to see the impact. An internal collection of a few hundred change agents, operating under the moniker “GM 2020,” was beginning to make its mark. The 70 executives who had been through the first two yearlong “Transformational Leaders” programs were spinning out new programs like Maven, a car-sharing service. At GM’s technology center in Warren, Mich., one floor had been cleared out for Maven. It had around a dozen cubicles, a few offices with doors, a carpeted area that was used for publicity and marketing, and a lot of empty space.

This spring I went back to Warren to see GM’s radical overhaul of the tech center. A half-hour drive north of GM headquarters in downtown Detroit, the tech center, designed by Eero Saarinen in the 1950s, is the real heart of the company. Twenty thousand engineers, designers, and IT specialists work here in 38 buildings.

“What you saw two years ago was just the beginning,” says chief of talent Michael Arena, an industrial engineer who studied human networks and innovation at MIT “Now,” he continues, “it’s in the blood of the company.”

Arena is describing the culture shift, but he might as well be referring to the building we are in, the Vehicle Engineering Center. The spacious lobby, an airy, two-story-high glass enclosure, features a Starbucks, a variety of comfortable couches, and the Trans Am from Smokey and the Bandit. We checked in at a surprisingly unobtrusive security desk and went to the second floor. “This is all about the work,” says Arena. “It’s all about creating spaces for the different ways people choose to work.”

Introverts and others in search of quiet are served by a smattering of Steelcase’s portable Brody cubicles. Group work is often done at rows of white desks set off by temporary walls with panels in cool grays and bright primary colors. There’s a lounge area with a large flat-screen TV, and lots of open space to encourage random interactions. The tabletops in the dining area are hammered hoods of old Buicks and Chevys. Everything is set back from the building’s glass exterior, creating 622-foot-long corridors that look out onto the historic campus, and the floors are flooded with natural light. I’m reminded of the mile-long paths along the curved glass of Apple’s new headquarters in Cupertino.

Indeed, nothing about the new Warren campus would look out of place at Google, Facebook, Herman Miller, or any other company we more readily recognize as innovative. Top executives visited the campuses of all those companies for inspiration on a redesign that would suit GM’s new workflows. “This isn’t about furniture,” Barra explains. “It’s about creating an environment for collaboration and giving people tools they need to work effectively. How can we make sure you really have a work environment that’s enabling and empowering, instead of constricting?”

巴拉的問題很合時宜。人們可能輕易會把這家公司想象成工業時代的遺跡，但她的通用汽車面貌絕非如此。77000名領薪員工中，近40%都在公司工作少于五年。“這部分人的加入是自下而上轉變的催化劑?！卑⒗锛{說。

年輕員工全新的需求和期望正協助巴拉在全公司上下發動思維的轉變。“他們引入的技術和價值，他們在社交媒體方面的專長，都在公司上下產生著深遠影響。”巴拉解釋道?！耙恍┣闆r下，實際上是由下至上的狀態?！?

巴拉也有方法自上而下撬動革新。她帶領全公司一同制定了七條行為準則，以引導組織上下所有人的做法——從中國的一位車間工人到巴西的一名營銷專員。為了保證“敢想敢做”（Be Bold）、“創新在當下”（Innovate Now）、“讓我來”（It’s on Me）以及其他口號不僅僅是口號，她與阿里納還有人事團隊協同整改員工培訓，使之側重于任務導向的設計思維，而少了說教。10月，她提出一項頗具勇氣和志向的目標，為所有人設定了幾個長期方針：零事故，零排放，零擁堵。

巴拉對這“三零”倡議充滿激情?！耙话俣嗄陙?，一直為人們提供交通便利，我們很自豪?！彼龑ξ艺f?！捌嚫淖兞巳藗兊纳罘绞?、工作地點，也促成了城市間緊密聯系。它帶來一種人們至今仍熱愛著的自由。那么，隨之而來的有什么？安全隱患、車禍、環境問題，還有堵車的煩惱。有了新技術，我們現在有能力解決這些問題。”

這一計劃也為通用汽車公司年輕一代人帶來了他們渴求的集體目標意識。“我們知道，如果員工的個人目標與團隊的目標不契合，他們在第一年離開的幾率就高出86%?！卑⒗锛{說。

當今企業管理方面值得注意的一個趨勢是，CEO必須親自管理人才：員工問題不能再“外包”給人事了。巴拉貫徹了這一點。她將自己的核心角色描述為“制定策略，管控風險，授權他人去執行，以及確保人員負起責任。”她將很大一部分時間花在最后兩點上。4月至6月下旬，她將以15或20人為一組，會面公司270位最高主管，討論他們在推動行為改變方面取得的進展。每個季度，她都會帶領14位核心主管進行短期出差，以“關注我們如何建立恰當程度的信任和對未來的共同目標，以及如何應對沖突”。她的人事理念不妨稱為“涓滴效應”：只要她與核心團隊之間保持公開互信的關系，這些團隊成員就可能努力與他們的直接下屬建立同樣的關系，然后一級一級往下。

但要讓這一切都不至落空，巴拉還必須做出清晰、艱難的決策，并實現成果。她也確實一直如此。4月，她解雇了凱迪拉克（Cadillac）總裁，原因是：“我們希望發展的步伐再快一些。”她說。但她最艱難的抉擇還是為精簡公司規模以期增加利潤而做出的一系列舉措。巴拉一反幾十年來公司對市場份額的渴求，關閉了在印度和南非的運營項目，并將歐寶（Opel）和沃克斯豪爾（Vauxhall）賣給了標致集團（Groupe PSA）——標致汽車（Peugeot）和雪鐵龍（Citro?n）的母公司。4月，她保留了通用汽車在韓國的項目，但那也是就繁雜的協議與該國多個勢力強大的工會重新談判后才做出的決定。

正是這些決策，加上公司SUV以及輕卡的市場需求，才為通用汽車這家又病又不靠譜、不得不靠政府救助的公司實現了健康發展和持續創收。2015年和2016年，通用汽車匯報利潤均超過90億美元。2017年，由于出售歐寶和沃克斯豪爾的相關費用一次性支付，再加上新稅法的影響，公司虧損39億美元。不過，許多分析人士相當看好通用汽車本年度的前景。前兩年北美地區毛利率均超過10%，而公司于各地的利潤率目前均在上漲。這樣的成就使公司有資格在下一波出行交通浪潮中爭搶頭籌?！拔磥砻總€季度都按照計劃做好工作，我們就能做到兩者（短期與長期）兼顧?！卑屠f。

斯坦福的奧萊理指出，一家舊時代的公司想要轉型以適應新的科技時代，必須做好三個方面：設想、孵化，以及規?；?。“每個人都能有所設想，”他對我說?！懊總€CEO都在創新方面投錢，這也很好。一些企業在孵化階段做得不錯。但難點在于規模化。當你必須將產業和產能從當前可盈利的業務中抽離出來，并投入到利潤有限、甚至可能蠶食當前業務的新項目中去的時候，才是真正見分曉的時刻。”

巴拉持相同觀點?！拔覀兊馁Y金配置框架表明我們要再次投資于企業，在長期范圍內帶動股東價值。我們的資產負債表將維持在投資級別。其他的資金，我們將適當回報給股東?！彼f?！暗滓闹е?，就是我們要為公司的長遠發展做新一輪投資?！?

通用汽車公司一方面在核心業務中不盈利的項目上削減開支，另一方面為研發自動駕駛的電動汽車提供充裕的資金支持。2016年初，公司投入5億美元用于Lyft項目——一項被認為比Uber更“友好”的拼車服務。兩個月后，通用汽車收購Cruise。去年10月，公司就收購了Strobe。這家11人的初創公司在做的激光探測與測量系統（lidar）能為自動駕駛的汽車分辨障礙物并判斷距離。該類系統和汽車蓄電池高昂的成本必須大幅下降，自動駕駛的汽車才能達到與當前汽車相近的售價。

巴拉和通用汽車公司總裁丹·阿曼一直謹慎對待Cruise，唯恐束縛其原有風格，或阻滯了它作為一家初創公司的發展速度。道格·帕克斯是雪佛蘭Bolt電動汽車研發工作的領導者，他負責監督通用汽車公司駕駛自動化的工作，并作為關鍵渠道連接底特律和舊金山。他的部分職責就是確保Cruise不受母公司侵害?！癈ruise的人可以來母公司，需要什么就拖走?！卑屠f?！暗腹静荒苋ツ沁叄堑栏瘛づ量怂怪纻€中原因。這樣，我們既能保證一家初創公司傳統的特點，又能讓母公司的支持成為他們強大的優勢。”

Cruise正在大規模招賢納士，而員工數已從被收購時的40人擴充至現在的500人。其中多數在舊金山一座平凡而不起眼的工業大樓里辦公，大樓前攔著一組鐵門，8英寸厚，而且看似有保安監控。然而，我造訪的時候，外面卻沒有一人執勤，內部也沒人回應對講機。我在想是否來錯了地方。當我看到停車場有幾輛白色雪佛蘭Bolt電動汽車，車上帶著Cruise標志和自動駕駛設備，終于有一名保安過來讓我進去。

Cruise確實給人以初創公司的感覺，有各種員工福利：開放空間、谷物食品自助機、鮮有拘束的工作氛圍。此外其發展速度看來也確實不愧為一家初創公司。短短兩年間，Cruise研發出了四款大幅升級后的Bolt自動駕駛電動汽車。最原始的版本中，Cruise的感應器和電腦由人工栓在車頂，而一些線路僅僅用膠帶固定，車也只有最基本的自動駕駛能力。但最新一款車在通用汽車公司一家工廠完成裝配：40%的原件都是Bolt自動駕駛汽車獨有的配置，此外也除去了方向盤以及其他人工操作。

Cruise和通用汽車公司的高管都認為如此高效的更新證明了兩家企業已成功融合。正如其他汽車制造商，通用汽車公司一般需要幾年才能研發一款新車，部分原因是每項系統的安全性能都需要反復驗證。一款車一旦投入生產，通常在至少一年內不會再去修改。然而，對于Cruise的軟件團隊，一天發布多項更新是常態。“你得想辦法把這兩個極端結合起來?！卑⒙谝淮坞娫挷稍L中對我說。“表面上看，兩個過程完全沒有交集，但實際上我們已經找到法子讓兩方面協同發展。我們造好了自動駕駛汽車的整體構架，使之能適應必要的更新節奏，不管是軟件的每日更新、一些部件的每月更新，還是其他部件永久不再更新?！?

Barra’s question is timely. It’s easy to think of the company as a relic of the industrial age, but her GM doesn’t look like that. Some 40% of its 77,000 salaried workers have been with the company less than five years. “That influx is the catalyst for change from below,” says Arena.

2

The different needs and expectations of that younger workforce are helping Barra drive a mindset change throughout the company. “The skills and assets they bring, their expertise in social media, all these things are having a real impact throughout the company,” Barra explains. “In some cases, there’s reverse mentoring going on.”

Barra also has levers she uses to drive change from above. She led a companywide effort to define a set of seven behaviors to inform the actions of anyone anywhere in the organization, from a factory worker in China to a marketer in Brazil. To ensure that “Be Bold,” “Innovate Now,” “It’s on Me,” and the others are more than catchphrases, she has worked with Arena and her HR team to shape training sessions that are heavy on task-oriented design thinking and light on exhortation. In October she unveiled an audacious aspirational goal to focus everyone on a set of long-term targets: zero accidents, zero emissions, and zero congestion.

Barra is passionate about the Zero Zero Zero initiative. “We’re very proud that we’ve been providing mobility to people for over 100 years,” she tells me. “Cars changed the way people lived, where they work, how cities came together. It gave people a freedom that they still love. Well, what came with that? Safety issues, crashes, environmental concerns, and the frustration of congestion. With the new technologies, we’re now equipped to solve those problems.”

The program also provides the kind of collective sense of purpose that GM’s younger generation yearns for. “We know that people are 86% more likely to leave in their first year if their purpose doesn’t align with the purpose of the organization,” says Arena.

A noteworthy trend in management circles these days is that CEOs must take charge of their company’s talent: They can’t outsource personnel issues to HR anymore. Barra embodies this. She describes her key roles as “creating strategy, managing risk, empowering people to execute, and making sure we hold people accountable.” She spends a remarkable amount of time on the last two. Between April and late July, she will meet with the company’s top 270 executives in groups of 15 or 20 people to discuss their progress at pushing behavioral change. Every quarter she leads her top team of 14 executives at offsites to “focus on how we build the right level of trust and a shared vision of the future, and on how we manage conflict.” She has what might be called a kind of trickle-down personnel theory: If she works to have an open, trusting relationship with her top team, then those people are likely to put in the effort to have the same with their direct reports, and so on down through the ranks.

None of this would matter if Barra didn’t also make clear, tough decisions and deliver results. But she does and has. In April she fired the president of the Cadillac division, because, she says, “We were looking to pick up the pace.” Her toughest decisions, however, have been a set of moves that shrank the company with the goal of increasing profitability. Reversing decades of hungry pursuit of market share, Barra shuttered operations in India and South Africa, and removed the company from Europe by selling Opel and Vauxhall to Groupe PSA, the parent company of Peugeot and Citro?n. In April she decided to keep GM in South Korea, but only after renegotiating onerous deals with the country’s powerful unions.

Those decisions, combined with demand for the company’s SUVs and light trucks, have turned a sick and unreliable company that the government had to save into a healthy, consistent earner. GM reported more than $9 billion in profits in both 2015 and 2016. For 2017 the automaker took a $3.9 billion loss, thanks to one-time charges related to the sale of Opel-Vauxhall and the new tax law. But many analysts are bullish on GM’s outlook for this year. Gross margins in North America exceeded 10% in each of the past two years, and margins are now rising across the company. That success has earned the company the right to contend for leadership in the next big wave of transportation. “Doing what we’ll say we’ll do quarter by quarter gives us the right to work on both [the short-term and the long-term],” says Barra.

Legacy companies trying to make the transition to a new technological era, argues Stanford’s O’Reilly, must succeed at three disciplines: ideation, incubation, and scale. “Everyone ideates,” he tells me. “Every CEO spends money on innovation, and that’s great. Some companies are okay at incubation. But scaling is the hard part. The rubber hits the road when you have to take assets and capabilities away from an existing profitable business and allocate them to new businesses that have lower margins and may cannibalize your existing business.”

Barra agrees. “Our capital allocation framework says we’re going to reinvest in the business to drive shareholder value over the long term, we’re going to maintain an investment grade balance sheet, and we’ll return the rest to shareholders in some fashion,” she says. “But the first pillar is that we’re going to reinvest in the business for the long term.”

As GM has cut back on unprofitable ventures in its core business, it has put real money behind its effort to develop an autonomous electric vehicle. In early 2016 the company made a $500 million investment in Lyft, the ride-sharing service known for being “nicer” than Uber. Two months later, GM acquired Cruise. Last October it bought Strobe, an 11-person startup that makes the “lidar” laser systems that help an autonomous car identify objects and discern its distance to them. The high cost of lidar systems and electric batteries will have to go way down for AVs to reach price points comparable today’s cars.

Barra and GM president Dan Ammann have been careful not to crimp Cruise’s style, or its startup speed. Doug Parks, who led the development of the electric Chevy Bolt, oversees GM’s autonomous efforts and serves as the critical conduit between Detroit and San Francisco. Part of his role is to protect Cruise from the mothership. “People at Cruise can go into the big company and pull out what they need,” Barra explains, “but the big company can’t come in unless Doug Parks knows what’s going on. It’s allowed us to keep the traditional things about a startup while giving them the great assets of a large company.”

Cruise has been on a hiring spree and now has more than 500 employees, up from 40 when it was acquired by GM. Most work from an unmarked, nondescript industrial building in San Francisco, behind a set of eight-inch-thick steel gates that are, in theory, monitored by security guards. When I arrive for a visit, however, no one is posted outside, and no one inside answers the intercom. I wonder if I’ve come to the right place. Just when I’m noticing that the parking lot has a couple of white Chevy Bolt EVs with the Cruise logo and autonomous hardware on them, a guard finally comes out to let me in.

Cruise does still feel like a startup, with all the trappings: open spaces, cereal dispensers, a visible informality. And it certainly seems to be moving at startup pace. In two years Cruise has developed four significantly upgraded versions of the autonomous Bolt EV. On the first version, the Cruise sensors and computers were bolted by hand to the roof, some wiring was simply taped into place, and the car had only rudimentary autonomous capabilities. The latest version, however, was built in a GM plant: 40% of its parts are unique to the autonomous Bolt, and it has no steering wheel or other human controls.

Both Cruise and GM executives point to this rapid iteration as proof that the two companies have successfully integrated. GM, like other auto manufacturers, typically takes years to develop a new car, in part because every system must be validated repeatedly for safety. When that car finally goes into production, nothing typically gets changed for at least a year. The Cruise software team, however, is accustomed to releasing new iterations multiple times a day. “Those are the two extremes you’re trying to bring together,” Ammann tells me in a phone interview. “On the surface, the processes seem completely orthogonal, but we have actually figured out how to have the two things work together. We’ve constructed the whole architecture of the AV to accommodate the necessary rates of change, whether it’s daily on software or monthly on some other components or never on others.”

同阿曼對話的時候，身處布魯克林（Brooklyn）的我正坐在停著的車內，等待傍晚六點整的到來——那時我就能把車停入泊位而不用收到紐約市警察局的罰單了。這種停車游戲就是傳統汽車帶來的諸多附屬效應中我相當樂意擺脫的一點。

去年11月，駕駛自動化方面的進展終于讓巴拉有足夠信心做出她迄今為止最大膽的聲明：通用汽車公司將在2019年以Cruise汽車的形式發布一款自動駕駛拼車服務，車輛沒有方向盤，也沒有其他人工操作。公司沒有透露該服務將在哪里發布，只說會在美國的某個城市。我問阿曼，Cruise正在繪制曼哈頓中心區域的地圖，發布地是否選在了紐約。比如在華爾街附近發布該項服務，或許能刺激公司的股票價格?！拔覀冞€沒有針對這個問題有所討論。”他卻說。阿曼是新西蘭裔，帶一種冷靜的幽默，他聽了會兒又繼續說道：“我倒是可以這么說。我們目前相當一部分研究都在舊金山開展，所以——就是說——可以算是一條線索。”

華爾街不厭其煩地追捧行車自動化的各項發布。原因又是什么？多數投資者關注的是長期回報，而在任何一家公司利用自動駕駛載客服務實現可觀營業額之前，我們還需等許多年，要創造利潤就更遙遠了。正如通用汽車公司，奧迪（Audi）、福特、大眾（Volkswagen）以及其他汽車生產商也做出了五花八門的承諾，要在某個日期之前將或這或那的自動駕駛服務投入使用?！斑@些汽車公司大都能‘達成’其目標，”邁克·拉姆齊說?！暗绊懣赡懿⒉淮蟆?/p>

說要在2021年之前推出完全自動駕駛的汽車?？扇f一最高時速只有40英里，并且只能在市中心行駛呢？”與我聊過的通用汽車公司以外的人當中，沒有一個相信這家公司能在明年發布任何有分量的相關產品，舊金山也好，其他地方也罷。

我將這些人的說法告訴巴拉，并問她是否擔心一旦自己的承諾落了空，公司將會受罰?！拔覀冋J為公司正走在正軌上，”她說?！拔蚁耄絹碓蕉嗟娜苏凉u漸明白，研發真正的無人駕駛汽車是相當困難的。我們的技術開發是走在正軌上的。車子的安全性能是我們的一大保障。并且，我們在做我們該做的。當然也有監管的層面，我們也在做相關工作。目前我們仍是不偏不倚，朝這個方向去做，我也相信我們能做到。這是我所要關注的。我不去想‘萬一不行怎么辦？’——我只關注怎樣才能把事情完成?！?

這樣謹慎地聲明了通用汽車公司的決心之后，她笑了笑?？偙粏柕轿磥碓鯓釉鯓樱彩窍喈斄钊似v的一件事吧，我想。

深刻的技術創新需要長時間孕育。所謂“服務型出行交通”（transportation as a service）的發展正處于最坎坷的階段。最終的愿景很明確，即在全球范圍內以安全、共享、自動駕駛的電動汽車取代當前的汽車，從而創造一個事故和污染更少的世界。但明確的目標并不意味著前方清晰的道路。當前，疑問和矛盾遠遠多于答案。找到這些答案所需要的時間越久，最終的愿景也就離我們越遠。

While Ammann and I talk, I’m sitting in my parked minivan in Brooklyn and waiting for 6 p.m. to arrive, so I can leave my vehicle in its spot without getting a ticket from the NYPD. The parking game is one of those collateral effects of traditional cars that I’ll be happy to give up.

By last November, Barra felt confident enough about the self-driving effort to make her boldest announcement yet: GM would launch an autonomous ride-sharing service featuring Cruise vehicles without steering wheels or other manual controls in 2019. The company didn’t disclose where the service would launch, other than allowing that it would be in a U.S. city. I ask Ammann if New York might be the site, since Cruise is currently mapping downtown Manhattan. A service in, say, the Wall Street area might juice the company’s stock price. “We haven’t commented on this,” he says. Ammann, a Kiwi with a dry sense of humor, pauses for a moment before continuing: “I will say this. We’re doing a significant majority of our testing in San Francisco, and so, you know, there’s a clue.”

Wall Street routinely cheers autonomous announcements. But why? Most investors are making long-term bets, and we are years away from anyone generating serious revenue by providing autonomous rides—and even further away from profits. Like GM has, Audi, Ford, Volkswagen, and other carmakers have made an array of promises about having one form of autonomy or another on the roads by a certain date. “Most of these car companies will ‘meet’ their goals,” says Mike Ramsey, “but the impact may not be great. Ford says they’ll have a fully autonomous vehicle by 2021. But what if the top speed is only 40 miles per hour and it’s limited to the urban core?” Nobody I spoke with outside of GM believes the company can launch anything of meaningful scale next year, in San Francisco or anywhere else.

I share with Barra what I’ve heard and ask if she worries that the company could be penalized for not delivering on her promise. “We believe we’re on track,” she says. “I think more and more people are understanding the difficulty of developing true autonomous vehicles that take the driver out. We are on track with our technology development. We will be gated by safety. And we’re going to do the right thing. Clearly, there’s regulatory aspects, and we’re working on those. We are still on track, we are still working to that, and I believe that we will. That’s where my focus is. I don’t focus on ‘What if we don’t?’—I’m focused on what will it take to do it.”

After that cautious affirmation of GM’s commitment, she smiles, and it occurs to me that it must be exhausting to have people constantly ask you to predict the future.

Radical new technologies take time to emerge. The development of what’s come to be called “transportation as a service” is in a particularly frustrating stage. The ultimate vision—a world where safe, shared, self-driving electric cars replace the ones we drive now, making the world safer and less polluted—is clear. But clarity about the destination doesn’t mean there’s an obvious path ahead. Right now, there are far more questions and contradictions than answers. The longer it takes to get those answers, the longer it will take to achieve the vision.

舉個例子，不少很有頭腦的人物都認為這將會帶來巨大的市場。Benchmark的比爾·古爾利是Uber早期的風險投資人，他曾告訴我：“這一市場可能比硅谷追逐過的任何市場都要巨大?！蓖ㄓ闷嚬镜陌⒙桶屠几嬖V我，這一發展“將改變世界”。Lyft的戰略部主任拉茲·卡普爾認為，光是邊緣效應就能瓦解價值5萬億美元的產業——他所指的包括汽車保險、貸款、加油站，甚至房地產。（卡普爾指出：“城市14%的房地產都是停車場。未來我們需要一些停車用地，但遠不需要這么多?！保?

但憑什么認為這些人就是正確的？目前電動汽車產業尚不具備有一定規模并且可持續的市場；Lyft和Uber都在拼命砸錢；而要保障自動駕駛汽車時刻都能安全、可靠地行駛于幾乎所有道路，還需要許多年的努力和幾十億美元的投資。“我們不了解的不是技術，”拉姆齊指出?！拔覀儾涣私獾氖沁@一切是否真正值得。”

于是就產生了這樣的疑問：如果一切努力都有了結果，市場價值又在哪里？Lyft和Uber都認為網絡是關鍵。我問卡普爾, Lyft是否像通用汽車公司一樣覺得前路充滿了各種不確定。“我們覺得是很明朗的，”他回答?！拔覀儾恍枰ネ品魏我粋€現有產業——比如一家傳統的汽車制造商。汽車銷量在減少，人們越來越將出行交通當作服務來消費，這一顯著變化已經被證實。我們以為，未來多少也是如此。在美國，當前全部汽車行駛里程中只有0.5%來自拼車公司。我們還有很大的成長空間。”

通用汽車公司和其他汽車制造商持不同意見?！罢l掌控了自動駕駛技術，掌控了研發的速度，誰就會處于一個非常微妙的地位，”阿曼說?！拔覀冋J為那將會是一個很重要的控制點?！?

這些截然相反的觀點也解釋了為什么通用汽車公司和Lyft之間的伙伴關系——借用一位內部人士的話——“萎縮”了。兩家公司已經不在任何項目上協作了。

伙伴關系確實不容易，但也越來越常見——部分原因是，當今的探索一旦產生什么新成果，各個企業就急迫渴望在相關領域站住腳跟?！盁o馬之車”還方興未艾的年代，就有幾百家團體搶著想把生意做起來。通用汽車公司本身就是在這趟風波中由多家企業合并而崛起的。

眼下這場私人出行方式新時代的爭霸賽中會出現怎樣的黑馬？有幾位競爭者的出現讓人倍感意外。德爾福集團（Delphi Automotive）曾是汽車制造企業的一家主要供應商，后來更名為安波福（Aptiv），開發自己的自動駕駛汽車。谷歌分出Waymo公司，其自動駕駛汽車已在加州行駛超過500萬英里。還有中國的百度——一家搜索引擎和科技公司，想要成為自動駕駛汽車界的“安卓”，并已同國內外130多家公司結成聯盟。

這一番排兵布陣大概只能算作一場小小的賽前切磋?！笆袌鰞r值的拔河比賽還沒真正開始，”阿曼說道?！暗鹊綗o人駕駛汽車即將投入使用，好戲才算上演。”

瑪麗·巴拉掌舵四年來，成功重塑了通用汽車公司。公司的核心業務蓬勃發展，有勢頭在未來幾年創造豐厚利潤——即使自動駕駛汽車登入各大城市早于預期。“我們的核心業務將在相當一段時間內保留其核心地位——向中美洲出售大型卡車，并在能耗和安全性能方面不斷優化。

通用汽車公司兼顧核心業務和邊緣項目，在兩方面都有不俗的發展速度。Bolt電動汽車從模型到生產只花了一年，初次亮相時，充一次電跑了238英里，表現優于預期。此外還有Cruise公司在自動駕駛方面的突破。“你不得不服。兩年半以前，沒人會談論通用汽車。這家公司完全無關痛癢，很落后?！绷硪患易詣玉{駛行業競爭者的某位主管對我說。

任職早期，巴拉讓所有高層主管做了MBTI（Myers-Briggs Type Indicator）性格測試。CEO本人是一位ISTJ，相對的人格類型就是“負責，真誠，善分析，務實，有序，勤奮，可信賴，在實際問題上能做出可靠判斷”?？磥?，通用汽車公司要把握“無馬之車”的未來，所需的正是這樣的品質。（財富中文網）

本文最先刊載于《財富》2018年6月1日刊。

譯者：沈昕宇

For example, many very smart people believe this is going to be a huge market. Bill Gurley of Benchmark, an early venture investor in Uber, once told me, “This is probably bigger than any market Silicon Valley has gone after.” GM’s Ammann and Barra both tell me it “will change the world.” Raj Kapoor, chief of strategy at Lyft, believes the side effects alone could disrupt $5 trillion of business—he’s thinking of such things as car insurance, loans, service stations, and even real estate. (Notes Kapoor: “14% of real estate in a city is parking. We’ll need some, but not nearly as much.”)

But what proof do we have that these people are right? There currently isn’t a viable mass market for electric vehicles; Lyft and Uber still lose money hand over fist; and it will take years and billions of dollars more investment to get self-driving cars to the point where they are safe and reliable on virtually all roads at all times. “What we don’t know isn’t the technology,” Ramsey points out. “What we don’t know is whether it will all be worth it.”

Then there’s this question: If all this effort does pay off, where will the real value lie? Lyft and Uber believe the network is what matters. I ask Kapoor whether Lyft, like GM, sees a lot of uncertainty ahead. “It feels really clear to us,” he says. “We don’t have an incumbent business that is about to be disrupted, like a traditional auto manufacturer. The big change that’s already been documented is people buying fewer cars and consuming transportation as a service. For us, the future is more of the same. In the U.S., just 0.5% of all the vehicle miles traveled are with ride-sharing companies. We have so much room to grow.”

GM and the automakers disagree. “The person who controls the autonomous technology, and the rate of development, will be in a very interesting position,” says Ammann. “We think that will be an important control point.”

Those diametrically opposed views help explain why GM’s partnership with Lyft has “dwindled,” in the words of one participant. The two companies are not working on anything together now.

Partnerships are really hard, but they are proliferating, in part because companies are desperate to secure some kind of foothold in whatever emerges from today’s explorations. In the early days of the horseless carriage, hundreds of outfits vied to establish themselves. General Motors itself rose from that morass, an amalgam of several companies.

What beast will come out of today’s scrum to battle for primacy in the new world of personal transportation? There are several unlikely contenders. Delphi Automotive, historically a key supplier to carmakers, has renamed itself Aptiv and is testing its own self-driving cars. Google has spun off Waymo, whose AVs have driven more than 5 million miles in California. Baidu, the Chinese search engine and technology company, wants to become the “Android” of autonomy and has formed an alliance with over 130 companies in China and around the world.

It’s likely that all this maneuvering is merely a very preliminary tussle. “The tug of war over value hasn’t started in earnest,” Ammann says. “This starts to get much more interesting when the driverless cars are about to begin operation.”

In four years at the helm, Mary Barra has remade GM. The core business is strong and in position to deliver solid profits for years to come, even if autonomous cars arrive in cities sooner than most people anticipate. “Our core business is going to be our core business for a long time,” she says. “Selling full-size trucks to Middle America, with increasingly better performance from a fuel-economy and safety perspective.”

GM is moving at an impressive pace, both in the core and on the edge of its business. The Bolt EV went from prototype to production in just one year. It debuted with a better-than-expected range of 238 miles per charge. And then there’s the autonomous progress at Cruise. “You’ve got to be impressed,” one executive at a self-driving competitor told me. “Two and a half years ago, GM was in no one’s conversation. They were totally irrelevant, a laggard.”

Early in her tenure, Barra asked all of her top execs to take the Myers-Briggs personality test. The CEO herself is an ISTJ, which the Myers-Briggs Type Indicator describes as, “Responsible, sincere, analytical, reserved, realistic, systematic. Hardworking and trustworthy with sound practical judgment.” It turns out that may be just what General Motors needs to figure out the future of horseless carriages.?

This article originally appeared in the June 1, 2018 issue of Fortune.